“Context – The smell of ecology.” ― R. Scott Bakker
Bidirectionality and Metaphor: An Introduction
Chanita Goodblatt Ben-Gurion University of the Negev, Department of Foreign Literatures and Linguistics Joseph Glicksohn Bar-Ilan University, Department of Criminology
Abstract The authors first present the interaction theory of metaphor, emphasizing its notion of bidirectionality. They then discuss the relationship between bidirectionality and blending, making explicit the different expectations regarding bidirectionality deriving from interaction theory and blending theory. With this as a suitable background for this special issue on bidirectionality in metaphor, the authors then provide a brief introduction to each of the essays that appear in the issue.
Keywords interaction theory, gestalt, blending, tension, cognition
The Interaction Theory of Metaphor It is not inconsequential that in developing his interaction theory of metaphor, Max Black uses the metaphor “man is a wolf,” a metaphor that is incompatible and even grotesque in its juxtaposition of man and animal. As we shall argue in our own essay in this issue, the grotesqueness of this Poetics Today 38:1 (February 2017) DOI 10.1215/03335372-3716189 q 2017 by Porter Institute for Poetics and Semiotics A preliminary version of this paper was presented at two venues: the research seminar held at Bar-Ilan University, in the Gonda Multidisciplinary Brain Research Center, entitled “The Creating Mind: Interdisciplinary Perspectives,” December, 2013; and at the third international conference on Cognitive Futures in the Humanities: Forging Futures from the Past, Worcester College, Oxford, April, 2015. We were supported by a Bar-Ilan University Faculty Research Grant, awarded to Joseph Glicksohn. Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 image is a key to understanding what we shall term a potential for “bidirectionality” in metaphor comprehension. Let us, however, develop this claim in stages. Max Black (1962: 41) writes: The effect … of (metaphorically) calling a man a “wolf ” is to evoke the wolfsystem of related commonplaces. If the man is a wolf, he preys upon other animals, is fierce, hungry, engaged in constant struggle, a scavenger, and so on. … The wolf-metaphor suppresses some details, emphasizes others — in short, organizes our view of man. … We can say that the principal subject is “seen through” the metaphorical expression — or, if we prefer, that the principal subject is “projected upon” the field of the subsidiary subject (emphasis original). If, at this preliminary stage, we look at the nominative “A is a B” metaphor, then following Black, we can attempt to visualize the process of metaphor comprehension (How is A a B?) by means of Figure 1. Much has been written on the analysis of such a nominative metaphor (e.g., Chiappe, Kennedy, and Chiappe 2003; Chiappe, Kennedy, and Smykowski 2003; Gentner et al. 2001; Glucksberg and Keysar 1990), yet little of it has explicitly adopted Black’s way of seeing the principal subject A (the tenor, Figure 1 Visualizing the process of metaphor comprehension 2 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 topic, target, primary subject) through the lens of the subsidiary subject B (the vehicle, source, secondary subject), or of “project[ing]” A “upon” the field of B. We have tried to capture this projection in Figure 1. As A is seen through the lens of B, A becomes much more similar to B than it previously was. This is a process, and not a static comparison. As Roger Tourangeau and Robert Sternberg (1982: 212 – 13) write, “Interaction theorists argue that the vehicle of a metaphor is a template for seeing the tenor in a new way. … In Black’s view … interpretation involves not so much comparing tenor and vehicle for existing similarities, as construing them in a new way so as to create similarity between them.” Black (1962: 44) continues, “If to call a man a wolf is to put him in a special light, we must not forget that the metaphor makes the wolf seem more human than he otherwise would.” In our understanding, and again with reference to Figure 1, it is true that, in the first stage of comprehending “A is a B,” A becomes much more similar to B than it previously was. But now B is “projected upon” the field of A, which has already undergone transformation in the first stage. So we have a new A and a new B, and these two are in juxtaposition in stage 2 of the process. Again, to cite Tourangeau and Sternberg (1982: 214), “metaphors generally involve seeing something (men) in one domain in terms of something (wolves) in a second domain, with a resulting change in our view of both domains.” And as Raymond Gibbs (1994: 238) stresses, “the whole point of interactionism is that both terms affect the meaning of the other. The ‘seeing as’ often associated with metaphor is multidirectional. If man is seen as wolf, so too is wolf seen as man in ‘Man is a wolf’ (Black, 1979, emphasis original).” This, then, is the interaction theory of Max Black, which is actually a theory about the process of metaphor comprehension. There are various stages in this process: in the first, A is seen through the lens of B; in the second, B is seen through the lens of an already transformed A; in the third, a transformed A is now seen through the lens of a transformed B, as well as a transformed B being seen through the lens of a transformed A. Clearly, then, A and B are shifting percepts or concepts within a dynamic, interactive process. Bidirectionality is an integral part of this interactive process because, while A is seen through the lens of B, B is also seen through the lens of an already transformed A. As Bipin Indurkhya has noted, in “the interaction, the target is structured in terms of the source, as much as this can be done because the target has its own attributes, which constrain how it can be structured, so that the resulting organization is influenced both by the source and the target” (2006: 144). Goodblatt and Glicksohn † Bidirectionality and Metaphor 3 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 Is there a resolution to this iterative process of metaphor comprehension? To address this question, we turn to Heinz Werner and Bernard Kaplan (1963: 21 – 22) who write: Here we come to an important tenet of organismic theory of symbol formation: the act of denotative reference does not merely, or mainly, operate with already formed expressive similarities between entities. Through its productive nature, it brings to the fore latent expressive qualities in both vehicular material and referent that will allow the establishment of semantic correspondence between the two entities. It is precisely this productive nature of the denotative act that renders possible a symbolic relation between any entity and another. Such a possibility could never be realized if one were dealing with static entities, namely, the symbolic vehicle as an end product and the referent as a preformed “thing out there.” It is only realized because it rests on twin form-building processes, one directed towards the establishment of meaningful objects (referents), the other directed towards the articulation of patterns expressive of meaning (vehicles) (emphasis original). As they insist, this is a dynamic, interactive, developmental process. Indeed, we suggest that if this process of metaphor comprehension is developmental in nature, then in line with Werner’s exposition of a general principle of development (1978: 108 – 9), such development “proceeds from a state of relative globality and lack of differentiation to a state of increasing differentiation, articulation, and hierarchic integration.” Specifically, the process of metaphor comprehension entails the differentiation and articulation of two unidirectional readings, namely, A is seen through the lens of B, and/or B is seen through the lens of A or, as we suggest, of an already transformed A. In addition, there is the possibility of their integration (what Werner terms “hierarchic integration”) that results in a higher-order gestalt. Indeed, as we have previously argued, metaphor is a gestalt (Glicksohn and Goodblatt 1993)—to which we now add that the process of metaphor comprehension might eventually lead to a higher-order gestalt, wherein the bidirectional readings are integrated. This, we stress, is a goal that is not necessarily achieved by readers of a text when interpreting poetic metaphor. Bidirectionality and Blending At this critical point in our presentation, it is necessary to distinguish this concept of bidirectionality from that of blending. To do so, we will first elaborate on the history of those theories of metaphor that are relevant for a discussion of both interaction and blending theories. Max Black’s interaction theory was heavily influenced by Ivor A. Richards’s proposal that “when we use a metaphor we have two thoughts of different things active 4 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 together and supported by a single word, or phrase, whose meaning is a resultant of their interaction” (1936: 93). Prior to Richards, there is a whole philosophical enterprise, including such scholars as Gustav Gerber (1820 – 1901), Philipp Wegener (1848 – 1916), Alfred Biese (1856 – 1930), and Karl Bu¨hler (1879 – 1963), paving the way for such a view of the interactive nature of metaphor (Nerlich and Clarke 2000, 2001). But it is Black (1962, 1979) who is credited with this notion of bidirectionality and of this explicit interaction theory. Yet George Lakoff and Mark Turner (1989: 131 – 32), in presenting their conceptual metaphor theory (CMT) as applied to poetic metaphor, were very thorough in their categorical dismissal of bidirectionality. They write: Unfortunately, this very real phenomenon has been analyzed incorrectly as follows: the target domain is described as “suffusing” the source domain, and it is claimed that the metaphor is bidirectional — from target to source as well as from source to target. Indeed, according to this theory, there is no source or target. There is only a connection across domains, with one concept seen through the filter of the other. Here’s what is wrong with such an analysis. When we understand that life is a journey we structure life in terms of a journey, and map onto the domain of life the inferential structure associated with journeys. But we do not map onto the domain of journeys the inferential structure associated with the domain of life. … We map one way only, from the source domain of journey onto the target domain of life. Yet even while interaction theory and its concept of bidirectionality is attacked, Black himself is not really mentioned; as Ray Jackendoff and David Aaron (1991: 322) note in their review of Lakoff and Turner’s book, “only in the appendix is Max Black cited as an adherent of this approach [interaction theory].” One plausible reason for ignoring Black is that he himself (1981) had reviewed the book Metaphors We Live By (1980), in which Lakoff and Mark Johnson first presented their conceptual metaphor theory, and was not forthcoming with praise. He writes, “Their exposition is endemically slipshod [and] the copious literature on metaphor is almost completely ignored” (1981: 210). Lakoff notes in an interview regarding his subsequent dismissal of Black, “I had read Black and I had no interest in what Black was doing. … What influenced me was the discovery that ordinary, everyday thought and language, and specially ordinary everyday thought, is structured metaphorically” (1998: 89). Subsequently, the line of scholarship developing prior to Richards (Nerlich and Clarke 2000, 2001) and then continuing from Richards and Black (Goodblatt and Glicksohn 2003, 2010) has been virtually sidestepped. CMT was Goodblatt and Glicksohn † Bidirectionality and Metaphor 5 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 presented as a new theory of metaphor, which dismisses the claims (or misunderstandings) of interaction theory (see, for example, Gibbs 1994: 238; Lakoff and Turner 1989: 133). CMT has itself been replaced by conceptual integration theory, or blending theory, wherein the interactive nature of metaphor has resurfaced as a “new” finding, demanding the modification of CMT, and indeed its evolution into blending theory. As Charles Forceville (2004: 86) notes in his review of this theory developed by Gilles Fauconnier and Mark Turner (2002), one of the alleged assets of blending theory that is repeatedly emphasized by Fauconnier and Turner is that it can explain emergent structure. Inasmuch as the blended space generates aspects of meaning that inheres in neither of the input spaces, conceptual integration yields something more than the sum of the component parts and hence clearly has a creative dimension. That is correct but, again, the notion of novel, emergent features has its roots in metaphor theory — not so much in the book on literary metaphor Turner himself coauthored with George Lakoff, More than Cool Reason (1989), but rather in Max Black’s “More about Metaphor” (1977/1979). Even now, this version of events is being revised, with Scarlet Marquette writing in a footnote that blending theory “is deeply indebted to Max Black, as well as to Black’s predecessors in the formulation of metaphor theory, I. A. Richards and Paul Ricoeur” (2007: 698). We propose that blending theory, with its emphasis on a process model for metaphor comprehension, has supplanted CMT, for essentially two reasons: the limitations of CMT’s concept of a unidirectional mapping, and the inability of CMT to convincingly address the emergent structure underlying poetic metaphor. As two of the major proponents of blending theory, Fauconnier and Turner (1998: 133 – 38) write: Conceptual integration — “blending” — is a general cognitive operation on a par with analogy, recursion, mental modeling, conceptual categorization, and framing. It serves a variety of cognitive purposes. It is dynamic, supple, and active in the moment of thinking. … In blending, structure from input mental spaces is projected to a separate, “blended” mental space. … The blend contains emergent structures not in the inputs. Furthermore, as elaborated by Joseph Grady, Todd Oakley, and Seana Coulson, “material is projected from both the source and target spaces to the blend. This arrangement contrasts with the simple, unidirectional projection posited by CMT, in which mappings are from source to target” (1999: 103). Thus, blending theory is quite sympathetic to the concept of bidirectionality. Even so, Lakoff (2014: 8), in discussing the recent work on bidirectionality in social cognition (Lee and Schwarz 2012), has recently cautioned 6 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 that “bidirectionality of experimental effect may or may not mean bidirectionality of the metaphorical mapping.” Be that as it may, we propose that while blending theory—with its emphasis on emergent structure and, further, of bidirectionality—might well contrast with CMT, it still presents a different theoretical position to metaphor comprehension in comparison with interaction theory. For while blending theory predicts the complete fusion of the two domains, we rather predict a continued potentiality for— and tension among—possible readings (Goodblatt and Glicksohn 2016a, 2016b), as we address in the next section. Bidirectionality and Metaphor The first point to note concerns the tension between the primary subject and the secondary subject of the metaphor. As Eva Kittay (1987: 184) has stressed, if we want to preserve the tension, we cannot give an account of interaction which neutralizes all tension between vehicle and topic … Unless this tension is preserved, the “suppress[ion of ] some details” and emphasis of others does not really organize our view of man, for unless the categories of man and wolf remain distinct we cannot use one distinct entity — with its systematic interconnections — to reconceive the other (emphasis original). This is the same point stressed by Richards (1936)—there is an inherent tension between the tenor and the vehicle of the metaphor. In other words, there cannot in principle be a total fusion of the two concepts; hence their blending can only be a potential that is not usually realized. Ray Jackendoff and David Aaron (1991: 334), in their review of Lakoff and Turner (1989), provide their own take on this interactive nature of metaphor: What is the outcome of creating a relationship between the incommensurable source and target domains? [Lakoff and Turner] claim that it is an understanding of the target domain in terms of the source domain. We suspect there is more, something like a “fusion” or “superimposition” of the source and target domains. … This hypothesis sharply differentiates metaphor from simile, where the source and target domains are merely compared, not superimposed. … According to this approach, the mental representation evoked by a metaphor, as well as its affective power, are the result of superimposing the meaning of the source and target domains. Fine details of a source image that do not find precise correlates in the target domain still contribute to the meaning and affect of the composite. Hence the proliferation of image detail in poetic metaphor is motivated: it contributes to the richness of the interpretation. Their comments on the inherent tension between the two subjects of the metaphor should be highlighted. They continue: Goodblatt and Glicksohn † Bidirectionality and Metaphor 7 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 In addition, the superimposition operation itself has important effects. The most obvious is the affect contributed by using one entity as a symbol for another. … A second effect of the superimposition operation is the sense of tension conveyed by incongruously fusing two disparate domains. The interpreter seeks to resolve this tension by finding points of contact or structural similarity between the two domains, so that they become point-by-point more congruent — this is the mapping process described by [Lakoff and Turner]. Yet this in turn can lead to a third effect, the production of further tension, as the domains themselves are refocused and restructured in order to bring about greater congruence — this is the “interaction” described by Black (1979). … (1991: 335) The second point, one that was raised by Forceville, is that “the ‘oscillation’ between the subjects may go on and lead to further elaborations of the metaphor” (1996: 21). In other words, as we have stressed here, we are referring to a process of metaphor comprehension, wherein one unidirectional reading may move into a second unidirectional reading, without necessarily achieving some form of resolution of the problem of how to reconcile one reading with the other. It is interesting to consider how such “oscillation” presents itself with respect to “man is a wolf.” As part of our own research, presented in this issue (Goodblatt and Glicksohn 2016b), we asked our participants to consider this metaphor prior to the reading of a poetic text. Table 1 presents two protocols as examples of such readings. Reader S5.S1 presents a unidirectional reading of man as wolf, first looking at physical similarity, but then refers to the wolf as “someone” who is “predatory like a person is predatory.” Reader S4.S1 sees them as echoing each other, with the “dehumanizing of the man” echoing “the personification of the wolf.” This is a fine bidirectional reading of the metaphor, and it echoes Black’s insight that Table 1 Protocols for the metaphor “man is a wolf” Reader S5.S1 Reader S4.S1 Man as Wolf Man as Wolf/Wolf as Man If man is a wolf then maybe looking like a wolf? Hair of a wolf, the eyes the ears and maybe the character of a wolf? A wolf is someone, a predator, sneaky, suspicious, but also loving and caring … a wolf is part of something like a man is part of something … predatory like a person is predatory … man is like a wolf in his character, his personality, but also can be in the notion of the man as a wolf, in the social notion. Because I see them [man, wolf ], it’s like they echo each other. The dehumanizing of the man echoes the personification of the wolf. 8 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 if “to call a man a wolf is to put him in a special light, we must not forget that the metaphor makes the wolf seem more human than he otherwise would” (1962: 44). We further claim that bidirectionality in a metaphor is sustained in two primary ways: a clash of sharp visual images and a use of the grotesque. If Gibbs (1994: 133) is correct in his assertion, that “imagery provides a means by which two previously dissimilar domains can be incorporated into one concept, because the task of comprehending metaphor presumably involves fusing two such domains,” then a clash of sharp visual images will prevent such fusion or blending. Furthermore, the emphasis here is on sharp because fuzzy images might be blended. The use of the grotesque, itself potentially involving such a clash of visual images (Harpham 1982: 11; Thomson 1972: 27), is a particularly interesting case, and we discuss this in our article in this special issue (Goodblatt and Glicksohn 2016b). This Special Issue on Bidirectionality The essays that we have commissioned for this special issue of Poetics Today bear on one or another aspect of bidirectionality in metaphor. We have ordered these essays such that a certain degree of balance is achieved both with respect to the disciplines represented and with respect to the claims made about bidirectionality in metaphor. In the first essay, the semiotician Marcel Danesi presents an overview of various theories of metaphor, suggesting that those implicating the interaction theory of metaphor, as explicated above, invoke notions of interaction, projection, and blending. And yet, as he writes, each model can be described as a “unidirectional” one, since it posits essentially that metaphor is the result of enlisting concrete vehicles in order to shed light on (and even construct) abstract topics. By and large, these models have not entertained the possibility that metaphor is actually a “bidirectional” process, whereby not only does it involve enlisting concrete vehicles to guide abstract conceptualization, but also the reverse — namely, that abstract topics allow us to understand the vehicles. In other words, the parts of a metaphor implicate each other in tandem. In their essay, the psycholinguists Albert N. Katz and Hamad Al-Azary suggest that three properties of semantic space “provide boundary conditions that invite uni- or bidirectionality when concepts are juxtaposed as in metaphor.” These properties are: (1) the distance of concepts A and B in this space; (2) the density of semantic space in which A and B reside; and (3) the semantic richness of this space for concrete and abstract concepts. Goodblatt and Glicksohn † Bidirectionality and Metaphor 9 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 They argue that “bidirectionality is more likely to be found when topic and vehicle come from semantically distant categories, in part because with distant categories there are fewer features that can be found relevant to both topic and vehicle.” Moving to linguistics, Roy Porat and Yeshayahu Shen address verbal metaphors and comparisons, and argue that the relation between the same two concepts/domains can, in principle, be either bidirectional or unidirectional. They suggest that bidirectionality is, in fact, “more basic than the unidirectional, in that it can be triggered by the mere presence of the two stimuli; in contrast, the unidirectional process requires an additional mechanism for it to be fully realized.” This additional mechanism is the linguistic structure of the verbal metaphor. For as they write: “In a sentence, the grammatical subject and the predicate typically tend to encode the target and the source, respectively.” Hence, unidirectionality stems from the structure of the metaphorical phrase. The cognitive psychologists David Anaki and Avishai Henik address the parallel between bidirectionality in metaphor and in synesthesia. They argue that while discussions of metaphors have suggested that metaphors are unidirectional and in particular move from the concrete to the abstract, current research has suggested that metaphors might work in the other direction also, namely, from abstract to concrete. Similarly, while most studies of synesthesia have documented its unidirectional nature, current research has provided evidence that synesthesia might be bidirectional. They address the question whether these similarities between synesthesia and metaphors are just superficial, or whether they tell us something about our cognitive mechanisms. The computer scientists Bipin Indurkhya and Amitash Ojha address advertisements in their essay, noting that while in “verbal metaphor, the target and the source domains can usually be distinguished clearly, and some features of the source domain are mapped to the target domain, and not vice versa,” this is far from clear in visual metaphor. They argue that “visual metaphors can appear to be symmetric more often than the verbal metaphors, because the lack of copula can turn the focus on the comparison between the source and the target, instead of the target itself.” The literary critic Margaret Freeman focuses in her essay on both Black’s Interaction Theory and Fauconnier and Turner’s Blending Theory. She concludes “that not only is metaphorical bidirectionality possible, it explains how the arts enable us to iconically connect with the world through our embodied cognition, not as objective observers in the Western classical sense, but as participatory sharers of that world.” Her examples range from a detailed exposition of “Man is a wolf,” discussed above, to the analysis of both advertisements and poetic texts. 10 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 In the final essay, Chanita Goodblatt (a literary critic) and Joseph Glicksohn (a cognitive psychologist) discuss the grotesque nature of John Donne’s poetic imagery, which constitutes “a clash of incompatibles, generated by the great distance between the two semantic fields.” We argue that it is this clash that sustains bidirectionality in a metaphor, by preserving the tension between its two subjects, while allowing each to alternatively become the focus of one’s attention while reading. Through both a cognitive-literary and an empirical study of the metaphors in Donne’s poems “The Bait” and “The Flea,” we show how these metaphors enable both embodied simulation and bodily feeling in the reader. We argue that Donne is, in fact, an early advocate of embodied cognition. The reader will thus find in this special issue an up-to-date assessment of the study of bidirectionality in metaphor. Our primary objectives are fivefold. First, we set out the intellectual history focused on the concept of bidirectionality, which is particularly evident in this “Introduction” and in the essay by Marcel Danesi. As these essays argue, the two major theories that promote the concept of bidirectionality in metaphor, the Interaction Theory of Metaphor and Blending Theory, both derive from a rich interdisciplinary heritage; we hope our readers will appreciate both the similarity—and especially the dissimilarity—between these two theories, and their predictions regarding bidirectionality. Secondly, as this special issue demonstrates, the interdisciplinary aspect of this intellectual history continues in current scholarship and research in metaphor. The essays in this issue present points of view from semiotics, psycholinguistics, linguistics, cognitive psychology, computer science and literary criticism, even while ultimately downplaying such disciplinary distinctions. Thus, for example, the next two essays are not entrenched in their respective disciplines of psycholinguistics (Katz and AlAzary) and linguistics (Porat and Shen), but rather bear as much relevance for the fields of cognitive psychology and for literary criticism as they do for the study of verbal metaphor on which the authors focus. Our third objective in editing this special issue is to highlight different methods of research: theoretical analyses, empirical studies, and textual analysis appear in various essays, and sometimes all within the same essay. This is clearly the case for the final three essays in this issue (Indurkhya and Ojha; Freeman; Goodblatt and Glicksohn); both theorists and those conducting empirical research in this domain should benefit from seeing how each type of scholarship can inform the other. Fourthly, visual metaphor, verbal metaphor and adjacent phenomena such as simile, synesthesia, and analogy, are all analyzed and demonstrated in the various essays. Thus, bidirectionality is not analyzed solely with respect to conceptual metaphor, and in turn with respect to Blending Theory, but also with respect to a wider frame of reference. Goodblatt and Glicksohn † Bidirectionality and Metaphor 11 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 What is more, researchers studying bidirectionality in metaphor should be acquainted with bidirectionality in synesthesia, as presented in the sixth essay in this issue (Anaki and Henik). At the same time, researchers studying synesthesia will now be able to view their work within this wider context of interest. Finally, our fifth goal in editing this special issue was to chart out paths for future research based on the scholarship presented here. Various ideas might come to mind on reading these essays: Can synesthetic metaphor inform our understanding of bidirectionality? If unidirectionality is a linguistic constraint of verbal metaphor, is bidirectionality encouraged by visual metaphor? What are the implications for the study of bidirectionality in metaphor for a specific poetic text—and a specific literary tradition? If our readers come away with an appreciation of the importance of bidirectionality in metaphor, as well as with questions they would like to investigate in their own work, then our objectives will have been fulfilled. References Anaki, David, and Avishai Henik 2016 “Bidirectionality in Synesthesia and Metaphor.” Special issue, Poetics Today 38, no. 1: 141 – 62. Black, Max 1962 Models and Metaphors: Studies in Language and Philosophy (Ithaca, NY: Cornell University Press). 1979 “More about Metaphor.” In Metaphor and Thought, edited by Anthony Ortony, 19 – 43 (Cambridge: Cambridge University Press). 1981 “Metaphors We Live By by George Lakoff, Mark Johnson” (Review), Journal of Aesthetics and Art Criticism 40, no. 2: 208 – 10. Chiappe, Dan L., John M. Kennedy, and Penny Chiappe 2003 “Aptness Is More Important than Comprehensibility in Preference for Metaphors and Similes,” Poetics 31, no. 1: 51 – 68. Chiappe, Dan, John M. Kennedy, and Tim Smykowski 2003 “Reversibility, Aptness, and the Conventionality of Metaphors and Similes,” Metaphor and Symbol 18, no. 2: 85 – 105. Danesi, Marcel 2016 “The Bidirectionality of Metaphor.” Special issue, Poetics Today 38, no. 1: 15 – 34. Fauconnier, Gilles, and Mark Turner 1998 “Conceptual Integration Networks,” Cognitive Science 22, no. 2: 133 – 87. 2002 The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities (New York: Basic Books). Forceville, Charles 1996 Pictorial Metaphor in Advertising (London: Routledge). 2004 “Book Review: The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities,” Metaphor and Symbol 19, no. 1: 83 – 89. Freeman, Margaret H. 2016 “Multimodalities of Metaphor: A Perspective from the Poetic Arts.” Special issue, Poetics Today 38, no. 1: 61 – 92. 12 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 Gentner, Dedre, Brian F. Bowdle, Phillip Wolff, and Consuelo Boronat 2001 “Metaphor Is Like Analogy.” In The Analogical Mind: Perspectives from Cognitive Science, edited by Dedre Gentner, Keith J. Holyoak, and Boicho N. Kokinov, 199 – 253 (Cambridge, MA: MIT Press). Gibbs, Raymond W. 1994 The Poetics of Mind: Figurative Thought, Language, and Understanding (New York: Cambridge University Press). Glicksohn, Joseph, and Chanita Goodblatt 1993 “Metaphor and Gestalt: Interaction Theory Revisited,” Poetics Today 14, no. 1: 83 – 97. Glucksberg, Sam, and Boaz Keysar 1990 “Understanding Metaphorical Comparisons: Beyond Similarity,” Psychological Review 97, no. 1: 3 – 18. Goodblatt, Chanita, and Joseph Glicksohn 2003 “From Practical Criticism to the Practice of Literary Criticism,” Poetics Today 24, no. 2: 207 – 36. 2010 “Conversations with I. A. Richards: The Renaissance in Cognitive Literary Studies,” Poetics Today 31, no. 3: 387 – 432. 2016a “Bidirectionality in Poetic Metaphor: William Carlos Williams and Imagist Poetry,” Versus: Quaderni di Studi Semiotici, 122. 2016b “Discordia Concors and Bidirectionality: Embodied Cognition in John Donne’s Songs and Sonnets.” Special issue, Poetics Today 38, no. 1: 163 – 88. Grady, Joseph E., Todd Oakley, and Seana Coulson 1999 “Blending and Metaphor.” In Metaphor in Cognitive Linguistics, edited by Raymond J. Gibbs Jr. and Gerard Steen, 101 – 24 (Amsterdam: John Benjamins). Harpham, Geoffrey Galt 1982 On the Grotesque: Strategies of Contradiction in Art and Literature (Princeton, NJ: Princeton University Press). Indurkhya, Bipin 2006 “Emergent Representations, Interaction Theory, and the Cognitive Force of Metaphor,” New Ideas in Psychology 24: 133 – 62. Indurkhya, Bipin, and Amitash Ojha 2016 “Interpreting Visual Metaphors: Asymmetry and Reversibility.” Special issue, Poetics Today 38, no. 1: 93 – 122. Jackendoff, Ray, and David Aaron 1991 “More than Cool Reason: A Field Guide to Poetic Metaphor by George Lakoff; Mark Turner,” Language 67, no. 2: 320 – 38. Katz, Albert N., and Hamad Al-Azary 2016 “Principles That Promote Bidirectionality in Verbal Metaphor.” Special issue, Poetics Today 38, no. 1: 35 – 60. Kittay, Eva F. 1987 Metaphor: Its Cognitive Force and Linguistic Structure (Oxford: Clarendon). Lakoff, George 1998 “Cognitive Semantics in the Heart of Language: An Interview with George Lakoff,” Fo´rum lingu¨ístico 1, no. 1: 83 – 119. 2014 “Mapping the Brain’s Metaphor Circuitry: Metaphorical Thought in Everyday Reason,” Frontiers in Human Neuroscience 8: article 958. http://dx.doi.org/10.3389/fnhum .2014.00958. Lakoff, George, and Mark H. Johnson 1980 Metaphors We Live By (Chicago: University of Chicago Press). Lakoff, George, and Mark Turner 1989 More than Cool Reason: A Field Guide to Poetic Metaphor (Chicago: University of Chicago Press). Goodblatt and Glicksohn † Bidirectionality and Metaphor 13 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021 Lee, Spike W. S., and Norbert Schwarz 2012 “Bidirectionality, Mediation, and Moderation of Metaphorical Effects: The Embodiment of Social Suspicion and Fishy Smells,” Journal of Personality and Social Psychology 103, no. 5: 737 – 49. Marquette, Scarlet 2007 “Metaphors We Lie by: Cognitive Blending in the Poetry of Elena Shvarts,” Slavic and East European Journal 51, no. 4: 693 – 715. Nerlich, Brigitte, and David D. Clarke 2000 “Blending the Past and the Present: Conceptual and Linguistic Integration, 1800 – 2000,” Logos and Language 1, no. 1: 3 – 18. 2001 “Mind, Meaning, and Metaphor: The Philosophy and Psychology of Metaphor in Nineteenth-Century Germany,” History of the Human Sciences 14, no. 2: 39 – 61. Porat, Roy, and Yeshayahu Shen 2016 “The Journey from Bidirectionality to Unidirectionality.” Special issue, Poetics Today 38, no. 1: 123 – 40. Richards, Ivor Armstrong 1936 The Philosophy of Rhetoric (New York: Oxford University Press). Thomson, Philip 1972 The Grotesque. The Critical Idiom 24 (London: Methuen). Tourangeau, Roger, and Robert J. Sternberg 1982 “Understanding and Appreciating Metaphors,” Cognition 1, no. 1: 203 – 44. Werner, Heinz 1978 “The Concept of Development from a Comparative and Organismic Point of View.” In Developmental Processes: Heinz Werner’s Selected Writings, edited by Sybil S. Barten and Margery B. Franklin, 107 – 30 (New York: International Universities Press). Werner, Heinz, and Bernard Kaplan 1963 Symbol Formation: An Organismic-Developmental Approach to Language and the Expression of Thought (New York: Wiley). 14 Poetics Today 38:1 Downloaded from http://read.dukeupress.edu/poetics-today/article-pdf/38/1/1/459694/POE381_01Goodblatt-Glicksohn_Fpp.pdf by guest on 21 September 2021
Models of Metaphor in NLP
Computer Laboratory University of Cambridge
15 JJ Thomson Avenue Cambridge CB3 0FD, UK Ekaterina.Shutova@cl.cam.ac.uk
Automatic processing of metaphor can be clearly divided into two subtasks: metaphor recognition (distinguishing be- tween literal and metaphorical language in a text) and metaphor interpretation (identifying the intended literal meaning of a metaphorical expression). Both of them have been repeatedly addressed in NLP. This paper is the first comprehensive and systematic review of the existing computational models of metaphor, the issues of metaphor annotation in corpora and the available resources.
Our production and comprehension of language is a multi-layered computational process. Humans carry out high-level semantic tasks effortlessly by subconsciously employing a vast inventory of complex linguistic devices, while simultaneously integrating their background knowledge, to reason about reality. An ideal model of language understanding would also be capable of performing such high-level semantic tasks.
However, a great deal of NLP research to date focuses on processing lower-level linguistic information, such as e.g. part-of-speech tagging, discovering syntactic structure of a sentence (parsing), coreference resolution, named entity recognition and many others. Another cohort of researchers set the goal of improving application- based statistical inference (e.g. for recognizing textual entailment or automatic summarization). In contrast, there have been fewer attempts to bring the state-of-the-art NLP technologies together to model the way humans use language to frame high-level reasoning processes, such as for example, creative thought.
The majority of computational approaches to
Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, pages 688–697, Uppsala, Sweden, 11-16 July 2010. ⃝c 2010 Association for Computational Linguistics
figurative language still exploit the ideas articulated three decades ago (Wilks, 1978; Lakoff and Johnson, 1980; Fass, 1991) and often rely on task-specific hand-coded knowledge. However, recent work on lexical semantics and lexical acquisition techniques opens many new avenues for creation of fully automated models for recognition and interpretation of figurative language. In this paper I will focus on the phenomenon of metaphor and describe the most prominent computational approaches to metaphor, as well the issues of resource creation and metaphor annotation.
Metaphors arise when one concept is viewed in terms of the properties of the other. In other words it is based on similarity between the concepts. Similarity is a kind of association implying the presence of characteristics in common. Here are some examples of metaphor.
- (1) Hillary brushed aside the accusations.
- (2) How can I kill a process? (Martin, 1988)
- (3) I invested myself fully in this relationship.
- (4) And then my heart with pleasure fills, And dances with the daffodils.1
In metaphorical expressions seemingly unrelated features of one concept are associated with another concept. In the example (2) the computational process is viewed as something alive and, therefore, its forced termination is associated with the act of killing.
Metaphorical expressions represent a great variety, ranging from conventional metaphors, which we reproduce and comprehend every day, e.g. those in (2) and (3), to poetic and largely novel ones, such as (4). The use of metaphor is ubiquitous in natural language text and it is a serious bottleneck in automatic text understanding.
1“I wandered lonely as a cloud”, William Wordsworth, 1804.
In order to estimate the frequency of the phenomenon, Shutova (2010) conducted a corpus study on a subset of the British National Corpus (BNC) (Burnard, 2007) representing various genres. They manually annotated metaphorical expressions in this data and found that 241 out of 761 sentences contained a metaphor. Due to such a high frequency of their use, a system capable of recognizing and interpreting metaphorical expressions in unrestricted text would become an invaluable component of any semantics-oriented NLP application.
Automatic processing of metaphor can be clearly divided into two subtasks: metaphor recognition (distinguishing between literal and metaphorical language in text) and metaphor interpretation (identifying the intended literal meaning of a metaphorical expression). Both of them have been repeatedly addressed in NLP.
2 Theoretical Background
Four different views on metaphor have been broadly discussed in linguistics and philosophy: the comparison view (Gentner, 1983), the interaction view (Black, 1962), (Hesse, 1966), the selectional restrictions violation view (Wilks, 1975; Wilks, 1978) and the conceptual metaphor view (Lakoff and Johnson, 1980)2. All of these approaches share the idea of an interconceptual mapping that underlies the production of metaphorical expressions. In other words, metaphor always involves two concepts or conceptual domains: the target (also called topic or tenor in the linguistics literature) and the source (or vehicle). Consider the examples in (5) and (6).
- (5) He shot down all of my arguments. (Lakoff and Johnson, 1980)
- (6) He attacked every weak point in my argument. (Lakoff and Johnson, 1980)
According to Lakoff and Johnson (1980), a mapping of a concept of argument to that of war is employed here. The argument, which is the target concept, is viewed in terms of a battle (or a war), the source concept. The existence of such a link allows us to talk about arguments using the war terminology, thus giving rise to a number of metaphors.
2 A detailed overview and criticism of these four views can be found in (Tourangeau and Sternberg, 1982).
However, Lakoff and Johnson do not discuss how metaphors can be recognized in the linguistic data, which is the primary task in the automatic processing of metaphor. Although humans are highly capable of producing and comprehending metaphorical expressions, the task of distinguishing between literal and nonliteral meanings and, therefore, identifying metaphor in text appears to be challenging. This is due to the variation in its use and external form, as well as a not clear-cut semantic distinction. Gibbs (1984) suggests that literal and figurative meanings are situated at the ends of a single continuum, along which metaphoricity and idiomaticity are spread. This makes demarcation of metaphorical and literal language fuzzy.
So far, the most influential account of metaphor recognition is that of Wilks (1978). According to Wilks, metaphors represent a violation of selectional restrictions in a given context. Selectional restrictions are the semantic constraints that a verb places onto its arguments. Consider the following example.
(7) My car drinks gasoline. (Wilks, 1978)
The verb drink normally takes an animate subject and a liquid object. Therefore, drink taking a car as a subject is an anomaly, which may in turn indicate the metaphorical use of drink.
3 Automatic Metaphor Recognition
One of the first attempts to identify and interpret metaphorical expressions in text automatically is the approach of Fass (1991). It originates in the work of Wilks (1978) and utilizes hand- coded knowledge. Fass (1991) developed a system called met*, capable of discriminating between literalness, metonymy, metaphor and anomaly. It does this in three stages. First, literalness is distinguished from non-literalness using selectional preference violation as an indicator. In the case that non-literalness is detected, the respective phrase is tested for being a metonymic relation us- ing hand-coded patterns (such as CONTAINER- for-CONTENT). If the system fails to recognize metonymy, it proceeds to search the knowledge base for a relevant analogy in order to discriminate metaphorical relations from anomalous ones. E.g., the sentence in (7) would be represented in this framework as (car,drink,gasoline), which does not satisfy the preference (animal, drink, liquid), as car
is not a hyponym of animal. met* then searches its knowledge base for a triple containing a hypernym of both the actual argument and the desired argument and finds (thing, use, energy source), which represents the metaphorical interpretation.
However, Fass himself indicated a problem with the selectional preference violation approach applied to metaphor recognition. The approach detects any kind of nonliteralness or anomaly in language (metaphors, metonymies and others), and not only metaphors, i.e., it overgenerates. The methods met* uses to differentiate between those are mainly based on hand-coded knowledge, which implies a number of limitations.
Another problem with this approach arises from the high conventionality of metaphor in language. This means that some metaphorical senses are very common. As a result the system would extract selectional preference distributions skewed towards such conventional metaphorical senses of the verb or one of its arguments. Therefore, although some expressions may be fully metaphorical in nature, no selectional preference violation can be detected in their use. Another counterargument is bound to the fact that interpretation is always context dependent, e.g. the phrase all men are animals can be used metaphorically, however, without any violation of selectional restrictions.
Goatly (1997) addresses the phenomenon of metaphor by identifying a set of linguistic cues indicating it. He gives examples of lexical patterns indicating the presence of a metaphorical expression, such as metaphorically speaking, utterly, completely, so to speak and, surprisingly, literally. Such cues would probably not be enough for metaphor extraction on their own, but could contribute to a more complex system.
The work of Peters and Peters (2000) concentrates on detecting figurative language in lexical resources. They mine WordNet (Fellbaum, 1998) for the examples of systematic polysemy, which allows to capture metonymic and metaphorical relations. The authors search for nodes that are relatively high up in the WordNet hierarchy and that share a set of common word forms among their descendants. Peters and Peters found that such nodes often happen to be in metonymic (e.g. publication publisher) or metaphorical (e.g. supporting structure – theory) relation
The CorMet system discussed in (Mason, 2004) is the first attempt to discover source-target do-
main mappings automatically. This is done by “finding systematic variations in domain-specific selectional preferences, which are inferred from large, dynamically mined Internet corpora”. For example, Mason collects texts from the LAB domain and the FINANCE domain, in both of which pour would be a characteristic verb. In the LAB domain pour has a strong selectional preference for objects of type liquid, whereas in the FINANCE domain it selects for money. From this Mason’s system infers the domain mapping FINANCE – LAB and the concept mapping money – liquid. He compares the output of his system against the Master Metaphor List (Lakoff et al., 1991) containing hand-crafted metaphorical mappings between concepts. Mason reports an accuracy of 77%, although it should be noted that as any evaluation that is done by hand it contains an element of subjectivity.
Birke and Sarkar (2006) present a sentence clustering approach for non-literal language recognition implemented in the TroFi system (Trope Finder). This idea originates from a similarity- based word sense disambiguation method developed by Karov and Edelman (1998). The method employs a set of seed sentences, where the senses are annotated; computes similarity between the sentence containing the word to be disambiguated and all of the seed sentences and selects the sense corresponding to the annotation in the most similar seed sentences. Birke and Sarkar (2006) adapt this algorithm to perform a two-way classification: literal vs. nonliteral, and they do not clearly de- fine the kinds of tropes they aim to discover. They attain a performance of 53.8% in terms of f-score.
The method of Gedigan et al. (2006) discriminates between literal and metaphorical use. They trained a maximum entropy classifier for this purpose. They obtained their data by extracting the lexical items whose frames are related to MO- TION and CURE from FrameNet (Fillmore et al., 2003). Then they searched the PropBank Wall Street Journal corpus (Kingsbury and Palmer, 2002) for sentences containing such lexical items and annotated them with respect to metaphoric- ity. They used PropBank annotation (arguments and their semantic types) as features to train the classifier and report an accuracy of 95.12%. This result is, however, only a little higher than the performance of the naive baseline assigning majority class to all instances (92.90%). These numbers
can be explained by the fact that 92.00% of the verbs of MOTION and CURE in the Wall Street Journal corpus are used metaphorically, thus making the dataset unbalanced with respect to the target categories and the task notably easier.
Both Birke and Sarkar (2006) and Gedigan et al. (2006) focus only on metaphors expressed by a verb. As opposed to that the approach of Krishnakumaran and Zhu (2007) deals with verbs, nouns and adjectives as parts of speech. They use hyponymy relation in WordNet and word bigram counts to predict metaphors at a sentence level. Given an IS-A metaphor (e.g. The world is a stage3) they verify if the two nouns involved are in hyponymy relation in WordNet, and if they are not then this sentence is tagged as containing a metaphor. Along with this they con- sider expressions containing a verb or an adjective used metaphorically (e.g. He planted good ideas in their minds or He has a fertile imagination). Hereby they calculate bigram probabilities of verb-noun and adjective-noun pairs (including the hyponyms/hypernyms of the noun in question). If the combination is not observed in the data with sufficient frequency, the system tags the sentence containing it as metaphorical. This idea is a modification of the selectional preference view of Wilks. However, by using bigram counts over verb-noun pairs Krishnakumaran and Zhu (2007) loose a great deal of information com- pared to a system extracting verb-object relations from parsed text. The authors evaluated their system on a set of example sentences compiled from the Master Metaphor List (Lakoff et al., 1991), whereby highly conventionalized metaphors (they call them dead metaphors) are taken to be negative examples. Thus they do not deal with literal examples as such: essentially, the distinction they are making is between the senses included in Word- Net, even if they are conventional metaphors, and those not included in WordNet.
4 Automatic Metaphor Interpretation
Almost simultaneously with the work of Fass (1991), Martin (1990) presents a Metaphor Interpretation, Denotation and Acquisition System (MIDAS). In this work Martin captures hierarchical organization of conventional metaphors. The idea behind this is that the more specific conventional metaphors descend from the general ones.
Given an example of a metaphorical expression, MIDAS searches its database for a corresponding metaphor that would explain the anomaly. If it does not find any, it abstracts from the example to more general concepts and repeats the search. If it finds a suitable general metaphor, it creates a mapping for its descendant, a more specific metaphor, based on this example. This is also how novel metaphors are acquired. MIDAS has been integrated with the Unix Consultant (UC), the system that answers users questions about Unix. The UC first tries to find a literal answer to the question. If it is not able to, it calls MIDAS which detects metaphorical expressions via selectional preference violation and searches its database for a metaphor explaining the anomaly in the question.
Another cohort of approaches relies on per- forming inferences about entities and events in the source and target domains for metaphor interpretation. These include the KARMA system (Narayanan, 1997; Narayanan, 1999; Feldman and Narayanan, 2004) and the ATT-Meta project (Barnden and Lee, 2002; Agerri et al., 2007). Within both systems the authors developed a metaphor-based reasoning framework in accordance with the theory of conceptual metaphor. The reasoning process relies on manually coded knowledge about the world and operates mainly in the source domain. The results are then projected onto the target domain using the conceptual mapping representation. The ATT-Meta project concerns metaphorical and metonymic description of mental states and reasoning about mental states using first order logic. Their system, however, does not take natural language sentences as input, but logical expressions that are representations of small discourse fragments. KARMA in turn deals with a broad range of abstract actions and events and takes parsed text as input.
Veale and Hao (2008) derive a “fluid knowledge representation for metaphor interpretation and generation”, called Talking Points. Talking Points are a set of characteristics of concepts belonging to source and target domains and related facts about the world which the authors acquire automatically from WordNet and from the web. Talking Points are then organized in Slipnet, a framework that allows for a number of insertions, deletions and substitutions in definitions of such characteristics in order to establish a connection between the target and the source
concepts. This work builds on the idea of slippage in knowledge representation for understanding analogies in abstract domains (Hofstadter and Mitchell, 1994; Hofstadter, 1995). Below is an example demonstrating how slippage operates to explain the metaphor Make-up is a Western burqa.
typically worn by women
expected to be worn by women must be worn by women
must be worn by Muslim women
By doing insertions and substitutions the system arrives from the definition typically worn by women to that of must be worn by Muslim women, and thus establishes a link between the concepts of make-up and burqa. Veale and Hao (2008), however, did not evaluate to which extent their knowledge base of Talking Points and the associated reasoning framework are useful to interpret metaphorical expressions occurring in text.
Shutova (2010) defines metaphor interpretation as a paraphrasing task and presents a method for deriving literal paraphrases for metaphorical expressions from the BNC. For example, for the metaphors in “All of this stirred an unfathomable excitement in her” or “a carelessly leaked report” their system produces interpretations “All of this provoked an unfathomable excitement in her” and “a carelessly disclosed report” respectively. They first apply a probabilistic model to rank all possible paraphrases for the metaphorical expression given the context; and then use automatically induced selectional preferences to discriminate between figurative and literal paraphrases. The selectional preference distribution is defined in terms of selectional association measure introduced by Resnik (1993) over the noun classes automatically produced by Sun and Korhonen (2009). Shutova (2010) tested their system only on metaphors expressed by a verb and report a paraphrasing accuracy of 0.81.
5 Metaphor Resources
Metaphor is a knowledge-hungry phenomenon. Hence there is a need for either an extenZsive manually-created knowledge-base or a robust knowledge acquisition system for interpretation of metaphorical expressions. The latter being a hard task, a great deal of metaphor research resorted to
the first option. Although hand-coded knowledge proved useful for metaphor interpretation (Fass, 1991; Martin, 1990), it should be noted that the systems utilizing it have a very limited coverage.
One of the first attempts to create a multipurpose knowledge base of source–target domain mappings is the Master Metaphor List (Lakoff et al., 1991). It includes a classification of metaphorical mappings (mainly those related to mind, feel- ings and emotions) with the corresponding examples of language use. This resource has been criticized for the lack of clear structuring principles of the mapping ontology (Lo ̈nneker-Rodman, 2008). The taxonomical levels are often confused, and the same classes are referred to by different class labels. This fact and the chosen data representation in the Master Metaphor List make it not suitable for computational use. However, both the idea of the list and its actual mappings ontology inspired the creation of other metaphor resources.
The most prominent of them are MetaBank (Martin, 1994) and the Mental Metaphor Data- bank4 created in the framework of the ATT-meta project (Barnden and Lee, 2002; Agerri et al., 2007). The MetaBank is a knowledge-base of English metaphorical conventions, represented in the form of metaphor maps (Martin, 1988) contain- ing detailed information about source-target concept mappings backed by empirical evidence. The ATT-meta project databank contains a large number of examples of metaphors of mind classified by source–target domain mappings taken from the Master Metaphor List.
Along with this it is worth mentioning metaphor resources in languages other than English. There has been a wealth of research on metaphor in Spanish, Chinese, Russian, German, French and Italian. The Hamburg Metaphor Database (Lo ̈nneker, 2004; Reining and Lo ̈nneker-Rodman, 2007) contains examples of metaphorical expressions in German and French, which are mapped to senses from EuroWordNet5 and annotated with source–target domain mappings taken from the Master Metaphor List.
Alonge and Castelli (2003) discuss how metaphors can be represented in ItalWordNet for
5EuroWordNet is a multilingual database with wordnets for several European languages (Dutch, Italian, Spanish, Ger- man, French, Czech and Estonian). The wordnets are structured in the same way as the Princeton WordNet for English. URL: http://www.illc.uva.nl/EuroWordNet/
Italian and motivate this by linguistic evidence. Encoding metaphorical information in general domain lexical resources for English, e.g. Word-Net (Lo ̈nneker and Eilts, 2004), would undoubtedly provide a new platform for experiments and enable researchers to directly compare their results.
6 Metaphor Annotation in Corpora
To reflect two distinct aspects of the phenomenon, metaphor annotation can be split into two stages: identifying metaphorical senses in text (akin word sense disambiguation) and annotating source – target domain mappings underlying the production of metaphorical expressions. Traditional approaches to metaphor annotation include manual search for lexical items used metaphorically (Pragglejaz Group, 2007), for source and target domain vocabulary (Deignan, 2006; Koivisto-Alanko and Tis- sari, 2006; Martin, 2006) or for linguistic mark- ers of metaphor (Goatly, 1997). Although there is a consensus in the research community that the phenomenon of metaphor is not restricted to similarity-based extensions of meanings of iso- lated words, but rather involves reconceptualization of a whole area of experience in terms of an- other, there still has been surprisingly little inter- est in annotation of cross-domain mappings. How- ever, a corpus annotated for conceptual mappings could provide a new starting point for both linguistic and cognitive experiments.
6.1 Metaphor and Polysemy
The theorists of metaphor distinguish between two kinds of metaphorical language: novel (or poetic) metaphors, that surprise our imagination, and conventionalized metaphors, that become a part of an ordinary discourse. “Metaphors begin their lives as novel poetic creations with marked rhetorical effects, whose comprehension requires a special imaginative leap. As time goes by, they become a part of general usage, their comprehension be- comes more automatic, and their rhetorical effect is dulled” (Nunberg, 1987). Following Orwell (1946) Nunberg calls such metaphors “dead” and claims that they are not psychologically distinct from literally-used terms.
This scheme demonstrates how metaphorical associations capture some generalisations govern- ing polysemy: over time some of the aspects of the target domain are added to the meaning of a
term in a source domain, resulting in a (metaphorical) sense extension of this term. Copestake and Briscoe (1995) discuss sense extension mainly based on metonymic examples and model the phenomenon using lexical rules encoding metonymic patterns. Along with this they suggest that similar mechanisms can be used to account for metaphoric processes, and the conceptual mappings encoded in the sense extension rules would define the lim- its to the possible shifts in meaning.
However, it is often unclear if a metaphorical instance is a case of broadening of the sense in context due to general vagueness in language, or it manifests a formation of a new distinct metaphorical sense. Consider the following examples.
b. (9) a. b.
As soon as I entered the room I noticed the difference.
How can I enter Emacs? My tea is cold.
He is such a cold person.
Enter in (8a) is defined as “to go or come into a place, building, room, etc.; to pass within the boundaries of a country, region, portion of space, medium, etc.”6 In (8b) this sense stretches to describe dealing with software, whereby COM- PUTER PROGRAMS are viewed as PHYSICAL SPACES. However, this extended sense of enter does not appear to be sufficiently distinct or conventional to be included into the dictionary, although this could happen over time.
The sentence (9a) exemplifies the basic sense of cold – “of a temperature sensibly lower than that of the living human body”, whereas cold in (9b) should be interpreted metaphorically as “void of ardour, warmth, or intensity of feeling; lacking enthusiasm, heartiness, or zeal; indifferent, apathetic”. These two senses are clearly linked via the metaphoric mapping between EMOTIONAL STATES and TEMPERATURES.
A number of metaphorical senses are included in WordNet, however without any accompanying semantic annotation.
Metaphor Identification Pragglejaz Procedure
Pragglejaz Group (2007) proposes a metaphor
identification procedure (MIP) within the frame-
6Sense definitions are taken from the Oxford English Dic- tionary.
work of the Metaphor in Discourse project (Steen, 2007). The procedure involves metaphor annotation at the word level as opposed to identifying metaphorical relations (between words) or source–target domain mappings (between concepts or do- mains). In order to discriminate between the verbs used metaphorically and literally the annotators are asked to follow the guidelines:
- Foreachverbestablishitsmeaningincontext and try to imagine a more basic meaning of this verb on other contexts. Basic meanings normally are: (1) more concrete; (2) related to bodily action; (3) more precise (as opposed to vague); (4) historically older.
- If you can establish the basic meaning that is distinct from the meaning of the verb in this context, the verb is likely to be used metaphorically.
Such annotation can be viewed as a form of word sense disambiguation with an emphasis on metaphoricity.
6.2.2 Source – Target Domain Vocabulary
Another popular method that has been used to ex- tract metaphors is searching for sentences containing lexical items from the source domain, the tar- get domain, or both (Stefanowitsch, 2006). This method requires exhaustive lists of source and target domain vocabulary.
Martin (2006) conducted a corpus study in order to confirm that metaphorical expressions occur in text in contexts containing such lex- ical items. He performed his analysis on the data from the Wall Street Journal (WSJ) cor- pus and focused on four conceptual metaphors that occur with considerable regularity in the corpus. These include NUMERICAL VALUE AS LOCATION, COMMERCIAL ACTIVITY AS CONTAINER, COMMERCIAL ACTIVITY AS PATH FOLLOWING and COMMERCIAL ACTIVITY AS WAR. Martin manually compiled the lists of terms characteristic for each domain by examining sampled metaphors of these types and then augmented them through the use of thesaurus. He then searched the WSJ for sentences containing vocabulary from these lists and checked whether they contain metaphors of the above types. The goal of this study was to evaluate predictive ability of contexts containing vocabulary from (1) source domain and (2) target
domain, as well as (3) estimating the likelihood of a metaphorical expression following another metaphorical expression described by the same mapping. He obtained the most positive results for metaphors of the type NUMERICAL-VALUE- AS-LOCATION (P Metaphor Source = ààà69, P M etaphor T arget = àà6àà, P M etaphor M etaphor = àààà3).
6.3 Annotating Source and Target Domains
Wallington et al. (2003) carried out a metaphor an- notation experiment in the framework of the ATT- Meta project. They employed two teams of an- notators. Team A was asked to annotate “interesting stretches”, whereby a phrase was considered interesting if (1) its significance in the document was non-physical, (2) it could have a physical significance in another context with a similar syntactic frame, (3) this physical significance was related to the abstract one. Team B had to annotate phrases according to their own intuitive definition of metaphor. Besides metaphorical expressions Wallington et al. (2003) attempted to annotate the involved source – target domain mappings. The annotators were given a set of mappings from the Master Metaphor List and were asked to assign the most suitable ones to the examples. However, the authors do not report the level of interannotator agreement nor the coverage of the mappings in the Master Metaphor List on their data.
Shutova and Teufel (2010) adopt a different approach to the annotation of source – target domain mappings. They do not rely on predefined mappings, but instead derive independent sets of most common source and target categories. They propose a two stage procedure, whereby the metaphorical expressions are first identified using MIP, and then the source domain (where the basic sense comes from) and the target domain (the given context) are selected from the lists of cate- gories. Shutova and Teufel (2010) report interannotator agreement of0.61( ).
7 Conclusion and Future Directions
The eighties and nineties provided us with a wealth of ideas on the structure and mechanisms of the phenomenon of metaphor. The approaches formulated back then are still highly influential, although their use of hand-coded knowledge is becoming increasingly less convincing. The last decade witnessed a high technological leap in
natural language computation, whereby manually crafted rules gradually give way to more robust corpus-based statistical methods. This is also the case for metaphor research. The latest develop- ments in the lexical acquisition technology will in the near future enable fully automated corpusbased processing of metaphor.
However, there is still a clear need in a unified metaphor annotation procedure and creation of a large publicly available metaphor corpus. Given such a resource the computational work on metaphor is likely to proceed along the following lines: (1) automatic acquisition of an extensive set of valid metaphorical associations from linguistic data via statistical pattern matching; (2) using the knowledge of these associations for metaphor recognition in the unseen unrestricted text and, finally, (3) interpretation of the identified metaphorical expressions by deriving the closest literal paraphrase (a representation that can be directly embedded in other NLP applications to enhance their performance
Besides making our thoughts more vivid and filling our communication with richer imagery, metaphors also play an important structural role in our cognition. Thus, one of the long term goals of metaphor research in NLP and AI would be to build a computational intelligence model accounting for the way metaphors organize our conceptual system, in terms of which we think and act.
I would like to thank Anna Korhonen and my reviewers for their most helpful feedback on this paper. The support of Cambridge Overseas Trust, who fully funds my studies, is gratefully acknowledged.
R. Agerri, J.A. Barnden, M.G. Lee, and A.M. Walling- ton. 2007. Metaphor, inference and domain- independent mappings. In Proceedings of RANLP- 2007, pages 17–23, Borovets, Bulgaria.
A. Alonge and M. Castelli. 2003. Encoding information on metaphoric expressions in WordNet-like resources. In Proceedings of the ACL 2003 Workshop on Lexicon and Figurative Language, pages 10–17.
J.A. Barnden and M.G. Lee. 2002. An artificial intelligence approach to metaphor understanding. Theoria et Historia Scientiarum, 6(1):399–412.
J. Birke and A. Sarkar. 2006. A clustering approach for the nearly unsupervised recognition of nonlit- eral language. In In Proceedings of EACL-06, pages 329–336.
M. Black. 1962. Models and Metaphors. Cornell Uni- versity Press.
L. Burnard. 2007. Reference Guide for the British Na- tional Corpus (XML Edition).
A. Copestake and T. Briscoe. 1995. Semi-productive polysemy and sense extension. Journal of Seman- tics, 12:15–67.
A. Deignan. 2006. The grammar of linguistic metaphors. In A. Stefanowitsch and S. T. Gries, editors, Corpus-Based Approaches to Metaphor and Metonymy, Berlin. Mouton de Gruyter.
D. Fass. 1991. met*: A method for discriminating metonymy and metaphor by computer. Computa- tional Linguistics, 17(1):49–90.
J. Feldman and S. Narayanan. 2004. Embodied mean- ing in a neural theory of language. Brain and Lan- guage, 89(2):385–392.
C. Fellbaum, editor. 1998. WordNet: An Electronic Lexical Database (ISBN: 0-262-06197-X). MIT Press, first edition.
C. J. Fillmore, C. R. Johnson, and M. R. L. Petruck. 2003. Background to FrameNet. International Journal of Lexicography, 16(3):235–250.
M. Gedigan, J. Bryant, S. Narayanan, and B. Ciric. 2006. Catching metaphors. In In Proceedings of the 3rd Workshop on Scalable Natural Language Un- derstanding, pages 41–48, New York.
D. Gentner. 1983. Structure mapping: A theoretical framework for analogy. Cognitive Science, 7:155– 170.
R. Gibbs. 1984. Literal meaning and psychological theory. Cognitive Science, 8:275–304.
A. Goatly. 1997. The Language of Metaphors. Rout- ledge, London.
M. Hesse. 1966. Models and Analogies in Science. Notre Dame University Press.
D. Hofstadter and M. Mitchell. 1994. The Copycat Project: A model of mental fluidity and analogy- making. In K.J. Holyoak and J. A. Barnden, editors, Advances in Connectionist and Neural Computation Theory, Ablex, New Jersey.
D. Hofstadter. 1995. Fluid Concepts and Creative Analogies: Computer Models of the Fundamental Mechanisms of Thought. HarperCollins Publishers.
Y. Karov and S. Edelman. 1998. Similarity-based word sense disambiguation. Computational Lin- guistics, 24(1):41–59.
P. Kingsbury and M. Palmer. 2002. From TreeBank to PropBank. In Proceedings of LREC-2002, Gran Canaria, Canary Islands, Spain.
S. Narayanan. 1999. Moving right along: A computa- tional model of metaphoric reasoning about events. In Proceedings of AAAI 99), pages 121–128, Or- lando, Florida.
G. Nunberg. 1987. Poetic and prosaic metaphors. In
Proceedings of the 1987 workshop on Theoretical issues in natural language processing, pages 198– 201.
G. Orwell. 1946. Politics and the english language. Horizon.
W. Peters and I. Peters. 2000. Lexicalised system- atic polysemy in wordnet. In Proceedings of LREC 2000, Athens.
Pragglejaz Group. 2007. MIP: A method for iden- tifying metaphorically used words in discourse. Metaphor and Symbol, 22:1–39.
A. Reining and B. Lo ̈nneker-Rodman. 2007. Corpus- driven metaphor harvesting. In Proceedings of the HLT/NAACL-07 Workshop on Computational Approaches to Figurative Language, pages 5–12, Rochester, New York.
P. Resnik. 1993. Selection and Information: A Class- based Approach to Lexical Relationships. Ph.D. the- sis, Philadelphia, PA, USA.
E. Shutova and S. Teufel. 2010. Metaphor corpus an- notated for source – target domain mappings. In Pro- ceedings of LREC 2010, Malta.
E. Shutova. 2010. Automatic metaphor interpretation as a paraphrasing task. In Proceedings of NAACL 2010, Los Angeles, USA.
G. J. Steen. 2007. Finding metaphor in discourse: Pragglejaz and beyond. Cultura, Lenguaje y Rep- resentacion / Culture, Language and Representation (CLR), Revista de Estudios Culturales de la Univer- sitat Jaume I, 5:9–26.
A. Stefanowitsch. 2006. Corpus-based approaches to metaphor and metonymy. In A. Stefanowitsch and S. T. Gries, editors, Corpus-Based Approaches to Metaphor and Metonymy, Berlin. Mouton de Gruyter.
L. Sun and A. Korhonen. 2009. Improving verb clus- tering with automatically acquired selectional pref- erences. In Proceedings of EMNLP 2009, pages 638–647, Singapore, August.
R. Tourangeau and R. Sternberg. 1982. Understand- ing and appreciating metaphors. Cognition, 11:203– 244.
T. Veale and Y. Hao. 2008. A fluid knowledge repre- sentation for understanding and generating creative metaphors. In Proceedings of COLING 2008, pages 945–952, Manchester, UK.
Koivisto-Alanko and H. Tissari. 2006. Sense and sensibility: Rational thought versus emotion in metaphorical language. In A. Stefanowitsch and S. T. Gries, editors, Corpus-Based Approaches to Metaphor and Metonymy, Berlin. Mouton de Gruyter.
S. Krishnakumaran and X. Zhu. 2007. Hunting elusive metaphors using lexical resources. In Proceedings of the Workshop on Computational Approaches to Figurative Language, pages 13–20, Rochester, NY.
G. Lakoff and M. Johnson. 1980. Metaphors We Live By. University of Chicago Press, Chicago.
G. Lakoff, J. Espenson, and A. Schwartz. 1991. The master metaphor list. Technical report, University of California at Berkeley.
B. Lo ̈nneker and C. Eilts. 2004. A Current Re- source and Future Perspectives for Enriching Word- Nets with Metaphor Information. In Proceedings of the Second International WordNet Conference— GWC 2004, pages 157–162, Brno, Czech Republic.
B. Lo ̈nneker-Rodman. 2008. The hamburg metaphor database project: issues in resource creation. Lan- guage Resources and Evaluation, 42(3):293–318.
B. Lo ̈nneker. 2004. Lexical databases as resources for linguistic creativity: Focus on metaphor. In Pro- ceedings of the LREC 2004 Workshop on Language Resources for Linguistic Creativity, pages 9–16, Lis- bon, Portugal.
J. H. Martin. 1988. Representing regularities in the metaphoric lexicon. In Proceedings of the 12th con- ference on Computational linguistics, pages 396– 401.
J. H. Martin.
Metaphor Interpretation. Academic Press Profes- sional, Inc., San Diego, CA, USA.
J. H. Martin. 1994. Metabank: A knowledge-base of metaphoric language conventions. Computational Intelligence, 10:134–149.
J. H. Martin. 2006. A corpus-based analysis of con- text effects on metaphor comprehension. In A. Ste- fanowitsch and S. T. Gries, editors, Corpus-Based Approaches to Metaphor and Metonymy, Berlin. Mouton de Gruyter.
Z. J. Mason. 2004. Cormet: a computational, corpus-based conventional metaphor extraction sys- tem. Computational Linguistics, 30(1):23–44.
S. Narayanan. 1997. Knowledge-based action repre- sentations for metaphor and aspect (karma. Tech- nical report, PhD thesis, University of California at Berkeley.
1990. A Computational Model of
A. M. Wallington, J. A. Barnden, P. Buchlovsky, L. Fel- lows, and S. R. Glasbey. 2003. Metaphor annota- tion: A systematic study. Technical report, School of Computer Science, The University of Birming- ham.
Y. Wilks. 1975. A preferential pattern-seeking seman- tics for natural language inference. Artificial Intelli- gence, 6:53–74.
Y. Wilks. 1978. Making preferences more active. Ar- tificial Intelligence, 11(3):197–223.